ITMI20010813A1 - EVALUATION OF LIVER FIBROSIS WITH ALGORITHMS BASED ON SERUM MARKERS - Google Patents

Description

Description of the patent for industrial invention entitled:

"Assessment of liver fibrosis score with serum marker-based algorithms"

Progressive liver fibrosis is a major cause of morbidity and. mortality in the world. Recent scientific advances show that the pathogenic process of fibrosis in the liver critically depends on the proliferation and activation of hepatic stellate cells (also called lipocytes, fat-storing cells or Ito cells) that synthesize and secrete extracellular matrix proteins (1). Furthermore, it is clear that this process is common to liver diseases of all etiologies. Of particular importance are chronic viral hepatitis B and C and alcoholic liver disease as well as genetic and autoimmune diseases of the liver, all of which lead to clinical problems through the common end route of progressive liver fibrosis, with the eventual development of cirrhosis. .

An important concept is the distinction between liver fibrosis and cirrhosis. Liver fibrosis is a reversible accumulation of the extracellular matrix in response to chronic damage in which the nodules have not yet developed, while cirrhosis involves an irreversible process, in which thick bands of matrix completely surround the parenchyma, forming nodules. Consequently, any therapy must be directed to patients with reversible disease (fibrosis), which requires early identification and control of those at risk (2).

The severity and progress of liver fibrosis are difficult to assess, with liver biopsy remaining the most reliable clinical method at the moment. The qualitative assessment of liver fibrosis by biopsy is limited by the variability between observers. Biopsies are clearly inadequate for the early clinical phase of drug development, where it is imperative to use less invasive methods to identify effective compounds in a commercially acceptable time pattern, usually measured in weeks to a maximum of 3 months of experimental therapeutic exposure. Further disadvantages are the low diagnostic specificity and bleeding risk. So there is a need for replacement markers of liver fibrosis. Serum tests allow non-invasive evaluation of fibrogenesis and fibrolysis in the liver and can be done repeatedly at short intervals (3). Serum tests that measure the dynamic processes of extracellular matrix synthesis (fibrogenesis) and degradation of the extracellular matrix (fibrolysis) reflect the amount of extracellular matrix present, the degree of fibrosis and the ongoing process of structural change in the liver (4 ).

The current state of the art of measuring liver fibrosis replacement markers is poorly developed. Previous studies have suggested that serum levels of extracellular matrix proteins (or their cleavage fragments) can be used to assess the severity and progression of liver fibrosis (4.5, US 5 316 914, EP 0283 779). Different serum markers were studied and correlations were found with liver biopsies and the severity of liver disease (6).

Some of the markers used for liver fibrosis assessment are PIIINP, Laminin, Hyaluronan, Collagen IV, TIMP-1, Tenascin, MMP-2, and Fibronectin. Markers were measured and their ability to assess liver fibrosis was demonstrated. However, the diagnostic value of each individual marker is not accurate and specific for evaluating fibrosis scores.

Then combinations of markers are studied to increase the diagnostic value; the simple biological index PGA which combines prothrombin time (PT), gamma-glutamyltranspeptidase (GGT) and apolipoprotein Al (Apo Al) and the PGAA index which includes alpha-2-macroglobulin (A2M). in the PGA index they have been described for the diagnosis of alcoholic liver disease in drinkers (7,8). Although the PGA and PGAA indices were combined with individual serum markers (9,10), the markers were not used to establish algorithms for assessing liver disease.

In this invention the serum markers of the extracellular matrix are brought together in a grouping that leads to a series of markers whose measurement allows the calculation of an algorithm and the use of this derived algorithm to predict the histological score of hepatic fibrosis. For this purpose, discriminant function analysis is used to determine which variables discriminate between the different fibrosis scores. The algorithms are derived from the marker series: N-terminal procollagen III propeptide (PIINP), Collagen IV, Collagen VI, Tenascin, Laminin, Hyaluronan, MMP-2, TIMP-1 and MMP-9 / TIMP-1 complex.

In general, all new techniques must be validated against existing standard techniques, if one exists. The current "gold standard" for evaluating liver fibrosis is liver biopsy. With the biopsy, some tissue randomly taken from the liver is cut into slices which are examined by an expert using a microscope. There are many problems associated with liver biopsies that induce some degree of uncertainty: the distribution of fibrosis in the liver (cluster fibrosis and the fact that the needle may have reached regions of the liver not affected by fibrosis), a preparation wrong sample (e.g. not enough tissue material) and the individuality and preferences of the pathologist (individual assessments). Furthermore, the fibrotic state of the liver is usually described using scores and there are many different and even incompatible scoring systems (eg Scheuer score, Ishak score, etc.).

For example, in a study with 24 patients, two independent pathologists had to evaluate the same biopsy samples for each patient at two different times using two different scoring systems. The number of evaluations in which the two pathologists came to identical results ranged from only 36% to 46%.

The new technique is based on the measurement of serum parameters, which are directly associated with the fibrotic process and on the combination of them at a mathematical level which leads to a fixed evaluation procedure.

To validate the new technique, the "gold standard" is not the best means but the only one, since a priori the two methods do not investigate comparable final values: while the serum parameters characterize dynamic processes, the biopsy characterizes the fibrotic manifestation at a fixed time . There may be a very active fibrotic process in the liver although the fibrotic tissue has not yet developed. Conversely, there may be large clusters of fibrotic tissue in the liver but the fibrotic activity may be stopped or temporarily stopped.

However, some mathematical functions of the serum parameters gave statistically different mean values significantly at different stages of the biopsy score. Discriminant analysis using the "gold-standard" was chosen to investigate the diagnostic power of these mathematical functions of the serum parameters.

Determination of serum marker concentration and subsequent calculation of an algorithm can be used to make decisions about whether to perform a biopsy or whether treatment should be started or continued or stopped. Hence the assignment of patients in. a group of biopsy scores without taking a biopsy is beneficial. The subdivision of patients into groups, for example light fibrosis compared to severe fibrosis, using algorithms is an advantage of the invention described.

Description of immumosaggi

The markers procollagen III propeptide (PIINP), Collagen IV, Collagen VI, Tenascin, Laminin, Hyaluronan, MMP-2, TIMP-1 and MMP-9 / TIMP-1 complex are used for the algorithms.

Markers are measured using sandwich immunoassays. The immunoassays of the invention comprise the reaction of two antibodies with human fluid samples, in which the capture antibody binds specifically to an epitope of the marker. The second antibody specific for a different epitope is used to detect this complex. Preferably the antibodies are monoclonal antibodies and both antibodies of said two test antibodies bind specifically to the protein.

Antibody or other similar terms used herein include both a complete monoclonal and polyclonal immunoglobulin as well as antigenic fragments or immunoreactive fragments that bind specifically to the marker, including Fab, .Fab, F (ab) 2 and F (v). The term antibody also includes binding proteins, specifically the hyaluronic acid binding protein (HABP).

The human fluid samples used in the assay of the invention can be any sample that contains the markers, e.g., blood, serum, plasma, urine, sputum or bronchoalveolar lavage (BAL). Typically a serum or plasma sample is used.

The antibodies of the invention can be prepared by techniques generally known in the art and are typically generated with respect to a sample of the markers.

The second antibody is conjugated to a detector group, e.g. alkaline phosphatase, horseradish peroxidase, or a fluorescent dye. The conjugates are prepared by techniques generally known in the art.

The concentration of markers in human fluids is measured and algorithms to assess the degree of fibrosis are calculated.

Statistical background

Discriminant function analysis is used to determine which variables discriminate between two or more naturally occurring groups. From the point of view of calculation it is very similar to the analysis of variance. The basic idea behind discriminant function analysis is to determine if groups differ with respect to the mean of a variable and then use that variable to predict membership of a group (eg, of new cases). Considered in this way, the problem of the function of the discriminant can be re-proposed as a one-way analysis of the variance problem (ANOVA). In particular, it may be asked whether two or more groups are significantly different from each other with respect to the mean of a particular variable. If the means for a variable are significantly different in different groups then we can say that this variable discriminates between groups. In the case of a single variable, the final test of significance of whether or not a variable discriminates between groups is the F test. F is essentially calculated as the ratio of the variance in the data between groups to the grouped (mean) variance within the group. . If the group variance is significantly greater then there must be significant differences between the means. Usually many variables are included in a study to see which one or which ones contribute to the discrimination between groups. In this case, we have a matrix of the total variances and co-variances; similarly we have a matrix of variances and co-variances grouped within a group. We can compare these two matrices through multivariate F tests to determine whether or not there are significant differences (with respect to all variables) between the groups. This is identical to the multivariate analysis of variance or MANOVA. As in MANOVA, one can first perform the multivariate test, and, if statistically significant, proceed to see which of the variables have significantly different means in the groups. So, even though the multiple-variable calculations are more complex, the main reasoning still applies, that is, that we are looking for variables that discriminate between groups, as evident in the observed mean differences.

For an observation group containing one or more quantitative variables and a classification variable that defines groups of observations, the discrimination procedure develops a discriminating criterion for classifying each observation into one of the groups. After that, predicting what happened in the past is not difficult. It is not unusual to obtain a very good classification if one uses the same cases from which the discriminating criterion was calculated. To get an idea of how well the current discriminating criterion works, different cases must be classified (a priori), that is, cases that have not been used to evaluate the discriminating criterion. Only the classification of new cases allows to evaluate the predictive validity of the discriminating criterion. To validate the derived criterion, the classification can be applied to other data groups. The dataset used to derive the discriminant criterion is called the "training" or calibration dataset.

The discriminating criterion (function or functions or algorithm) is determined through a measurement of the generalized square distance. It can be based on a pooled co-variance matrix that generates a linear function. Either Euclidean or Mahalanobis distance can be used to determine proximity.

For the development of the discriminating algorithm, data from a group of subjects from a study based on the observation of liver fibrosis were analyzed. The liver fibrosis scoring systems under consideration were:

------

Applying a step-by-step discriminant analysis, for example, the following serum parameter functions were shown to have a greater impact on the corresponding type of score.

A corresponding discriminant analysis gave the

linear discriminant functions that can be used for the calculation and prediction of the biopsy score. The algorithms can be applied to any known scoring system (e.g. Scheuer score, Ishak score, Netavir score, Ludwig score, HAI score).

Algorithms can be used to predict a patient's biopsy score (e.g.

score 0, 1, 2, 3, ...) or to predict a group of scores (category) to which a patient belongs (e.g. mild fibrosis; score from 0 to 1).

The discriminant functions used include combinations of markers chosen from the N-terminal procollagen III propeptide list (PIIINP), Collagen IV, Collagen VI, Tenascin, Laminin, Hyaluronan, MMP-2, TIMP-1 and MMP-9 / TIMP- complex. 1 and also factors between -1000 and 1000.

Different scores need a different algorithm from the list of markers and factors.

Examples

Example 1

Scheuer scoring algorithms.

The following algorithms 1, 2 and 3 were calculated by correlating biopsies evaluated with the Scheuer scoring system and serum marker concentrations from a group of patients with liver disease:

TO

[0

[

[2

[

[

The algorithms were used to predict biopsy scores from a separate group of patients. The calculated scores were compared with the scores determined by a single pathologist (case B), with a consensus score of 3 pathologists (case C), and with the range covered by all pathologists (case A). Kappa values, negative predictive values (NPV) for a score of 0-1, and positive predictive values (PPV) for a score of 2-4, for sensitivity and specificity were also calculated.

Example 2

Ishak scoring algorithms.

The following algorithms 1, 2 and 3 were calculated by correlating the biopsies evaluated with the Ishak scoring system and the serum marker concentrations of a group of patients with liver disease: Algorithm 1

The algorithms were used to predict biopsy scores from a separate group of patients. The calculated scores were compared with the scores

determined by a single pathologist (case B), with a consensus score of 3 pathologists (case C) and with the range covered by all pathologists (case A). Kappa values, negative predictive values (NPV) for a score of 0-2, positive predictive values (PPV) for a score of 3-6, sensitivity and specificity were calculated.

Example 3

Receiver Operating Characteristic (ROC) Curves for the Scheuer Score.

By grouping patients into categories of no / mild fibrosis (score 0-1) and moderate / severe fibrosis (score 2-4) for the Scheuer score and calculating the algorithms for a dichotomous result, the following results were obtained:

The algorithms were used to calculate the receiver operating characteristic curves for the categories with absent / mild fibrosis (score 0-1) and moderate / severe fibrosis (score 2-4) for the Scheuer score. The calculated scores were compared with the scores determined by a single pathologist (case B), with a consensus score of 3 pathologists (case C) and with the range covered by all pathologists (case A). The area under the curve (AUC) values were calculated.

Example 4

Receiver Operating Characteristic (ROC) Curves for the Ishak Score.

By grouping patients into categories with no / mild fibrosis (score 0-2) and moderate / severe fibrosis (score 3-6) for the Ishak score and calculating the algorithms for the dichotomous result, the following results were obtained:

The algorithms were used to calculate receiver operating characteristic curves for the categories with absent / mild fibrosis (score 0-2) and moderate / severe fibrosis (score 3-6) for the Ishak score. The calculated scores were compared with the scores determined by a single pathologist (ca.

so Β), with a consensus score of 3 pathologists (case C) and with the range covered by all pathologists (case A). The values of the curve under the area (AUC) were calculated.

LITERATURE

1. Friedman SL

The cellular basis of hepatic fibrosis Mechanism and treatment strategies.

N Engl J Med 1993; 328: 1828-1835

2. Friedman SL

Molecular mechanism of hepatic fibrosis and principle of therapy

J Gastroenterol 1997; 32: 424-430

3. Haysaka A, Saisho H

Serum markers as tools to monitor liver fibrosis Digestion 1998; 59: 381-384

4. Schuppan D, Stolzel U, Oesterling C, Somasundaram R

Serum assays for liver fibrosis

J Hepatol 1995; 22 (Suppl 2): 82-88

5. Murawaki Y, Ikuta Y, Nishimura Y, Koda M, Kawasa

Claims (10)

CLAIMS 1. Method for diagnosing liver fibrosis in which two or more diagnostic markers are measured in a body fluid and the measurements are combined through mathematical algorithms. 2. A method according to claim 1 where the algorithms are characterized in such a way that the diagnostic markers are selected in the N-terminal propeptide group of procollagen III (PIIINP), Collagen IV, Tenascin, Laminin, Hyaluronan, MMP-2, TIMP- 1 and MMP-9 / TIMP-1 complex. The method according to claim 1 wherein the body fluid is blood, serum, plasma or urine. The method according to claims 1 and 2 wherein the algorithms can be used to support, predict or replace the histological score of a liver biopsy. The method according to claim 4 where the algorithms can be adjusted to fit a known scoring system (e.g. Scheuer score, Ishak score, HAI score, Ludwig score, Metavir score). The method according to claim 1 and 2 wherein the algorithms can be used to divide patients into score groups. A method according to claim 4 and 6 wherein the treatment decisions can be supported by the calculated score or the calculated score group. The method according to claim 4 and 6 wherein the clinical trial of the drug for liver disease including the studies to determine the dose can be supported by the calculated score or the calculated score group. The method according to claim 4 and 6 wherein the decisions whether to perform a biopsy or not can be supported by the score calculated from the calculated score group. The method according to claim 6 where patients can be classified as suffering from absent / mild or moderate / severe liver fibrosis.